- * Title: - *
- * SequenceIdMatcher - *- * Description: - *
- * Routine which does approximate Sequence Id resolution by name using string + * Routines for approximate Sequence Id resolution by name using string * containment (on word boundaries) rather than equivalence. It also attempts to * resolve ties where no exact match is available by picking the the id closest * to the query. - *- * Copyright: Copyright (c) 2004 - *
- * - *- * Company: Dundee University - *
- * - * @author not attributable - * @version 1.0 */ public class SequenceIdMatcher { @@ -53,7 +37,8 @@ public class SequenceIdMatcher names = new Hashtable(); for (int i = 0; i < seqs.length; i++) { - // TODO: deal with ID collisions - SequenceI should be appended to list associated with this key. + // TODO: deal with ID collisions - SequenceI should be appended to list + // associated with this key. names.put(new SeqIdName(seqs[i].getDisplayId(true)), seqs[i]); // add in any interesting identifiers if (seqs[i].getDBRef() != null) @@ -84,9 +69,10 @@ public class SequenceIdMatcher */ private SequenceI pickbestMatch(SeqIdName candName, Vector matches) { - SequenceI[] st= pickbestMatches(candName, matches); - return st==null || st.length==0 ? null : st[0]; + SequenceI[] st = pickbestMatches(candName, matches); + return st == null || st.length == 0 ? null : st[0]; } + /** * returns the closest SequenceI in matches to SeqIdName and returns all the * matches to the names hash. @@ -95,11 +81,12 @@ public class SequenceIdMatcher * SeqIdName * @param matches * Vector of SequenceI objects - * @return Object[] { SequenceI closest SequenceI to SeqIdName, SequenceI[] ties } + * @return Object[] { SequenceI closest SequenceI to SeqIdName, SequenceI[] + * ties } */ private SequenceI[] pickbestMatches(SeqIdName candName, Vector matches) { - ArrayList best=new ArrayList(); + ArrayList best = new ArrayList(); SequenceI match = null; if (candName == null || matches == null || matches.size() == 0) { @@ -117,9 +104,10 @@ public class SequenceIdMatcher SequenceI cand = (SequenceI) matches.elementAt(0); matches.remove(0); names.put(new SeqIdName(cand.getName()), cand); - int q,w,candlen = cand.getName().length(); + int q, w, candlen = cand.getName().length(); // keep the one with an id 'closer' to the given seqnam string - if ((q=Math.abs(matchlen - namlen)) > (w=Math.abs(candlen - namlen)) + if ((q = Math.abs(matchlen - namlen)) > (w = Math.abs(candlen + - namlen)) && candlen > matchlen) { best.clear(); @@ -127,13 +115,17 @@ public class SequenceIdMatcher matchlen = candlen; best.add(match); } - if (q==w && candlen==matchlen) + if (q == w && candlen == matchlen) { // record any ties best.add(cand); } } - if (best.size()==0) { return null; }; + if (best.size() == 0) + { + return null; + } + ; return (SequenceI[]) best.toArray(new SequenceI[0]); } @@ -157,8 +149,10 @@ public class SequenceIdMatcher } /** - * Find all matches for a given sequence name. - * @param seqnam string to query Matcher with. + * Find all matches for a given sequence name. + * + * @param seqnam + * string to query Matcher with. */ public SequenceI[] findAllIdMatches(String seqnam) { @@ -221,6 +215,7 @@ public class SequenceIdMatcher } return pickbestMatch(nam, matches); } + /** * core findIdMatch search method for finding all equivalent matches * @@ -236,11 +231,10 @@ public class SequenceIdMatcher { matches.addElement(names.remove(nam)); } - SequenceI[] r=pickbestMatches(nam, matches); + SequenceI[] r = pickbestMatches(nam, matches); return r; } - private class SeqIdName { String id; @@ -286,7 +280,8 @@ public class SequenceIdMatcher * arbritrarily extended sequence id's (like portions of an aligned set of * repeats from one sequence) */ - private String WORD_SEP = "~. |#\\/<>!\""+((char)0x00A4)+"$%^*)}[@',?_"; + private String WORD_SEP = "~. |#\\/<>!\"" + ((char) 0x00A4) + + "$%^*)}[@',?_"; /** * matches if one ID properly contains another at a whitespace boundary. @@ -299,7 +294,8 @@ public class SequenceIdMatcher */ public boolean equals(SeqIdName s) { - // TODO: JAL-732 patch for cases when name includes a list of IDs, and the match contains one ID flanked + // TODO: JAL-732 patch for cases when name includes a list of IDs, and the + // match contains one ID flanked if (id.length() > s.id.length()) { return id.startsWith(s.id) ? (WORD_SEP.indexOf(id.charAt(s.id